AI & Academic Integrity

Introduction 

AI is developing rapidly and is now also affecting higher education more and more. AI support will be part of everyone's everyday life and there are many possibilities but also risks involved. Recently, information has spread about the ChatGPT robot, that can answer advanced questions and quickly create well-formulated texts based on various requests. (However, there are a number of other bots such as Google's Bard, Perplexity and others). 

What is truly new about this? 

A home exam where the students have to write a PM or similar has never been completely legally secure. Letting others (friends, "kind" parents or commercial players) write the student's essays has been happening for a very long time. Already in Linnaeus' time, you could buy a written thesis. 

However, it has now become fast, easy and accessible for everyone to be able to get a relatively well-written text in seconds. In addition, several of these chat bots are currently being integrated for example into Office 365, Microsoft BING and Google's search tools. 

Are our plagiarism check systems not working? 

No, existing plagiarism check systems like Ouriginal cannot find AI texts! Every time a person uses a chatbo, original texts are written, so there is no plagiarism in the usual sense. However, software that can recognize AI-written text is coming. However, there are always ways to get around technology-assisted control systems, so we probably should not rely exclusively on them. 

How should we deal with this? 

AI is here to stay so the issue needs serious management. Our point of view is that it is best handled through good information work (for both teachers and students), review of the university's policies and guidelines, and above all through well-thought-out examination strategies. Remember that AI-supported writing will be part of the students' future professional life, so we must also learn to handle it in a sensible way and see how we can use and integrate the technology in teaching without it leading to cheating. 

How will AI and chatbots affect the way we teach? 

Below you will find some things you can do right now!

• Inform yourselves about the possibilities of AI!
• Emphasize the importance of evaluation of sources!
• What do you wants to assess and how are you going to do it? Is it the knowledge connected to the subject or is it the ability to write?
• If it is the students' ability to write well-structured texts using correct language, it may be wise to use classroom exams where there is no access to the internet or AI support.
• Use oral examination more.
• Try combining different examination methods such as PM writing and oral examination
• Let the students be co-creators of the exam - the students develop and write questions, answers and assessment criteria. The teacher then selects what to use at the exam.
• Try to shift the focus from summative to formative assessment, i.e. follow the students' learning process and assess not only the end result but milestones in the process towards the learning objectives!
• Use part-time seminars where the students can describe how they intended to structure the material and how they went about finding their sources. Let them present their half-finished essays to the teacher and their classmates, etc.
• Look for changed language in the essays. Is this the normal way of expression?
• Be sure to follow up on students' references and make sure they also include page references in their citations. Not all chatbots provide references. Perplexity as an example often provide very simple references without page references. It also sometimes generates pure nonsense texts.
• Let the writing assignments connect to other parts of the course such as excursions, labs, seminars and lectures. The more specific details about a specific course required in the task, the more difficult it will be for the AI ​​tools to write a good text.
• Ask the students to be more personal in their writing or, for example, write logbooks about their learning and the tasks in the course. Allowing students to adopt a "meta-perspective" on their learning can be difficult for an AI to handle. For example, what has the student learned during the course, and how will this knowledge help them in their future professions?
• Finally, try to integrate AI into teaching in a constructive way! Among other things, students can practice looking for sources that support the factual background in an AI-written text and thereby increase their information literacy.

Watch the AI seminar from 230503:
AI & AI (Artificial Intelligence and Academic Integrity)

Associate professor Sonja Bjelobaba is researcher at the Centre for Research Ethics & Bioethics as well as an associate professor at the Department of Modern Languages the Uppsala university.